1778
J. Org. Chem. 1993,58,1778-1784
Ion Pair Carbon Acidities of Some Silanes in Tetrahydrofuran' Andrew Streitwieser,' Linfeng Xie, Peng Wang, and Steven M. Bachrach Department of Chemistry, University of California, Berkeley, California 94720 Received April 12,1990 (Revised Manuacript Received November 23,1992)
The relative solvent-separated ion pair (SSIP)lithium acidity (~KUITHF) and contact ion pair (CIP) cesium acidity (~KC~ITHF) were obtained for 9-fluorenyltrimethylsilane(1) (21.3,21.6,respectively) and 9-fluorenyl-tert-butyldimethylsilane(2)(20.3,20.6,respectively) in THF. Values for PKC~ITHF were determined at 25 "C for @-biphenylylmethy1)-tert-butyldimethylsilane(3),35.4,benzyltrim(6), 33.5, ethylsilane (4), 37.5,a,a-bis(trimethylsily1)toluene (5),34.1,2-(trimethylsilyl)-1,3-dithiane (trimethylsily1)acetonitrile(7),28.8,and tris(trimethylsily1)methane (8),36.8.Some thermodynamic parameters were determined by measurements at other temperatures,and some ionic acidities (pK(F1)) were determined by conductivity studies. Carbanion stabilization by these silyl substituents varies from about 1 to over 3 pK units in different systems. 9,9-Bis(trimethylsilyl)fluorene (9) was found to undergo silyl transfer on treatment with various carbanions,but this reaction is slower than proton transfer. Silicon-stabilized carbanions have become versatile intermediates in organic synthesise2The earlier example of their use in the Peterson olefin synthesis3has recently been supplemented by many additional reactions using silicon-stabilizedcarbanions.4 Several methods are available to generate these anions but proton abstraction by a lithium base provides a useful and common method. Kinetic and equilibrium acidity factors are important in the relative ease of deprotonation of such silanes. It has been well-documented that the trialkylsilyl group can behave as a weak electron acceptor when directly linked to a conjugated system. Examples are acidity measurementa of trimethylsilyl-substituted anilinium ions and phenols: infrared intensity studies of phenyltrimethylsilane? and hydrogen isotope exchange rates of n- and p(trimethylsi1yl)tolueneswith lithium cyclohexylamide in cyclohexylamine.' In recent years, there has been increasing interest in the effect of silyl groups on carbon acidity. Ab initio MO calculationson silyl-substitutedcarbanions have been used to investigate the effect,8t9 but the fiit experimental quantitative measurements of the silyl group stabilization effect on carbanions were done by Petrov et al., who reported the pK,'s of (trimethylsily1)acetylene and (triethylsily1)acetylenein DME with lithium as the gegenion (1) Carbon Acidity. 77. (2) (a) Colvin, E. Silicon in Organic Synthesis, 2d ed.;Butterworth London, 1988. (b) Chvalowky, V; Bellama, J. M. Carbon-finctionul Orgonosilicon Compounds; Plenum: New York and London, 1983. (c) Fritz, G.; Matem, E. Carboeilaner;Springer-Verlag: Heidelberg, 1986. (d) Patai, S.; Rappoport, Z., Eds. The Chemistry of Organic Silicon Compounds; Wiley: New York, 1989. (3) Peterson, D. J. J. Org. Chem. 1968, 33, 780. (4) (a) King, R. B.; Oliver, J. P. Organometallic Chemistry Reuiews; Annual Surveys: Silicon-Lead; Elsevier: Oxford, 1984. (b) Kanemaaa, S.; Tanaka, J.; Nagahama, H.; Tsuge, 0. Chem. Lett. 1985, 12223. (c) Block, E.; Anlam, M. Tetrahedron 1988, 44, 281. (d) Furin, G. G.; Vyazankina, 0. A.; Gosteveky, B. A.; Vyazankin, N. S. Ibid. 2675. (5) (a) Benkeeer, R. A.; Krysiak, H. R. J. Am. Chem. SOC.1968, 76, 2421. (b)Buell, G. R.; Corriu, R.; Guerin, C.; Spialter, L. J. Am. Chem. SOC.1970,92,7424. (6)Cutres3,N. C.; Katritzky, A. R.; Eabom, C.; Walton, D. R. M.; T o w n , R. D. J. Organomet. Chem. 1972,43,131. (7) Ma", F.; Streitwieser,A., Jr. J. Am. Chem. SOC.1967,89,3770. (8) HopLinaon,A. C.;Lien, M.H. J. Organomet. Chem. 1981,2M, 287. (9) Schleyer, P. v. R.; Clark, T.; Kos, A. J.; Spitmagel,G. W.; Rohde, C.; Arad, D.; Houk, K. N.; Rondan, N. G. J. Am. Chem. SOC.1984,106, 6467.
0022-3263/93/1958 1778$O4.00/0
to be 24.6 and 24.9,'O respectively, on a scale in which the pK of tert-butylacetylene is 26.0; thus, the effect of the silicon group in this system is about 1 pK unit. More recently, Miah and Fraser reported equilibrium lithium ion pair acidities of benzyltrimethylsilanell and [(phenylthi~)methyl]trimethylsilane~~ to be greater than that of toluene by NMR measurements with lithiated tetramethylpiperidine. Eaborn et al. found that tria(trimethylsily1)methane-t undergoes tritium exchange in DMSO/HzO/KOH about 5-7 times faster than triphenylmethane.l3 From the experimental electron affinity of (trimethylsily1)methyl radical. Wetzel and Brauman" have estimatedthat silicon stabilizesan alkyl anion by 20 kcal mol-' in the gas phase. Bordwell has reported ionic pKs for several silyl-substituted carbon acids in DMSO:169-fluorenyltrimethylsilane, 21.7(1.2pKunits per hydrogen more acidicthanfluorene), (trimethylsily1)malonicester, 19.0(2.3 pK units less acidic than malonic ester but 1.5 pK units more acidic than isopropylmalonic ester), and (phenylsulfony1)methylltriphenylsilane, 21.3 (7.7 pK units lower than phenyl methyl sulfone). These results indicatethat the carbanionstabilizing effect of a silyl substituent is highly variable. To provide further quantitative understanding of the effect of silyl substitution on carbon acidity and the effect of different metal counterions we determined the acidity in THF of the following compounds: 9-fluorenyltrime(2), @thylsilane (l), 9-fluorenyl-tert-butyldimethylsilane biphenylylmethy1)-tert-butyldimethylsilane(31,benzyltrimethylsilane (4)and a,a-bis(trimethylsily1)toluene(5). The lithium salts (1Land 2L)were shown to be solvent separated ion pairs (SSIP), and the cesium salts are undoubtedly contact ion pairs (CIP). The results are (10) (a)Petrov, E. S.;Terekhova,M.I.; shatemhtein,A. I. Zh. Obshch. Khim. 1974,44, 1118(Engl.Tranel.p 1076). (b)Petrov,E. S.;Tereknova, M.I.; Shatemhtein,A. I.; Tromdmov,B. A.; Mirekov, R. G.; Voronkov, M.G. Dokl. Akad. Nauk. SSSR. 1979,211 (Engl. Tranel. p 692). (11) Miah, M.A. J.; Frawr, R. R. J. Bangladesh Chem. SOC.1989,2, 77. (12) Miah,M. A. J.; Frawr, R. R. Ind. J. Chem. ISSO, B A , 688. (13) Eaborn, C.; Eidemhink, R.; Jackeon, P. M.;Walton, D. R. M.J. Organomet. Chem. 1976,101, C40. (14) Wetzel, D. M.; Brauman, J. I. J . Am. Chem. SOC.1988,110,8399. See alro: Damrauer, €2.; K m , S. R.; DePuy, C. H. Organometallics 1988, 7, 637. (16) Bordwell, F. G. Acc. Chem. Res. 1988,21,456.
Q 1993 American Chemical Society
J. Org. Chem., Vol. 58, No.7,1993 1779
Ion Pair Carbon Acidities of Some Silanes
Table 11. Relative Ion Pair Aciditiem of 1-4 with Indicaton in THF at 26
Table I. Spectral Data for the Saltr of 1-4 in THF at 26 O C ____
1L 2L 1c 2c 3c
4c 5c
364 363 356 356 472 372 373
RH
12OOO 12 OOO 9760 9760 43 800 19 210 18300
Results and Discussion
Ion Pair Acidities. The types of ion pairs involved were determined by spectroscopic analysis. 9-Lithio-9(trimethylsily1)fluorene(1L)and 9-lithh9-(tert-butyldimethyleily1)fluorene(2L)in THF have visible spectra similar to that of unsubstituted 9-lithiofluorene. The s'A, of 1Land 2L are at 364and 363nm,respectively. The shapes of the spectra do not depend on the concentrations of the salts, and Amax's do not changewith decreasein temperature to-30 OC. Addition of 15-crown-5,a powerful coordinating The agent for lithium cation, does not change the. ,A Am's of the corresponding cesium salts (1Cand 2C)are both 356 nm, a decrease of 8 nm relative to the lithium salts. The spectra are similarly insensitive to changes in salt concentration and temperature. As has been observed generally for fluorenyl anions, solvent-separated ion pairs (SSIP) have longer A, than the corresponding contact ion pairs (CIP).20 It is clear by analogy that the lithium salts 1L and 2L are SSIPs and the cesium salts 1C and 2C are CIPs. The cesium salts 3C and 4CZ1 are also assumed to be CIP. Thermodynamic and conductivity measurements of the lithium and cesium salts provide further confirmation of these ion pair assignments and will be discussed later. Spectral data for the study are presented in Table I. The accuracy of the extinction coefficient values for all compounds (direct titration) except 4is estimated as i 5 % . The extinction coefficient of 4 was measured by an alternative method (see Experimental Section), and the uncertainty is estimated to be 5 4 % . The Am's of the ~~
* * *
expressed relative to our ion pair acidity scales161g as PKLi/THF(SSIP) and PKC~/THF(CIP). Some ion pair dissociation constants were determined by conductivity measurements to give the correspondingrelative free ion acidities,pK(F1). Concentrationmeasurementsand thermodynamic data further confirmed the pK assignments. In the course of these measurements we observed some silylgrouptransfer reactions. Silylgroup transfer is slower than proton transfer and is much slower for tert-butyldimethylailyl than for trimethylsilyl. Additional measurementswere made on compoundsexpectedto give more localized carbanions: 2-(trimethylsilyl)-1,3-dithiane (6), (trimethylsily1)acetonitrile(71, and tris(trimethylsily1)methane (8).
(16) Bora, D. A.; Knufman, M.J.; Streitwieser,A.,
M+ Li+
InHb APKMITHF PKMITHF PKMITHF~V BA 1.18 0.02 21.31 21.3 BnMP 0.01 & 0.03 21.36 Cs+ 9-MeFI -0.75i 0.03 21.59 21.6 4 3 - m -1.22 & 0.06 21.69 2 Li+ BA 0.11 & 0.01 20.24 20.3 BnMP -1.03 0.06 20.32 CS+ 9-BnFI -0.73 0.02 20.57 20.6 9-MeFI -1.69 & 0.10 20.63 4.5-MP -2.31 & 0.11 20.60 3 Cs+ DPM 2.13 & 0.06 35.38 35.4 DoTM 1.15 i 0.01 35.37 35.34 DmXM -0.62 0.01 4 CS+ g M B -1.3 37.5 37.6 PpTS 0.4 37.6 6 Cs+ DPM 0.88 34.1 34.1 DoTM -0.06 34.2 6 Cs+ DPM 0.2 33.5 33.6 DoTM 4 . 7 33.5 7 cs+ 9-PSX 1.0 28.9 28.8 DP3 0.8 28.7 8 Cs+ PpTS -0.2 37.0 36.8 DmXM 0.7 36.7 a Per hydrogen. Abbreviations and pK's of indicators: lithium salts,17BA, benzanthrene, 20.13;BnMP,9-benzylbenzo[defJfluorene, 21.35; cesium 9-BnFl, 9-benzylfluorene, 21.30; g-MeFl, 9-methylfluorene,22.32; 4,5-MP, 4,5-methylenephenanthrene,22.91; DPM, diphenylmethane, 33.25; DoTM,di-o-tolylmethane, 34.22; DmXM, di-m-xylylmethane,36.96, p-MB, p-methylbiphenyl, 38.73; PpT& phenylp-tolyl sulfide,37.2, and PSX, 9-(phenylthio)xnuthene, 27.88 (to be 1
~
Jr. J. Am. Chem.
SOC.1985,107,6975. (17)Gronert. S.; Streitwieser, A., Jr. J. Am. Chem. SOC.1986, 108,
7016.
(18)Kaufmnn, M.J.; Gronert, S.;Streitwieser,A., Jr. J. Am. Chem. SOC.1988,110,2829. (19) Streitwieaer,A.; Ciuln,J. C.; &om, J. A.;Thiele,G.J. Org. Chem.
1991,56,1074-6. (20) Hogen-Each, T. E.; Smid, J. J. Am. Chem. Soc. 1966,88,307. Szwarc,M.Ions and Ion Pairs in Organic Reactions;John Wiley & SOM: New York, 1974. (21) O'Brien, D. H.; Russell,C. R.;Hart, A. J. J.Am. Chem. SOC.1979, 101,633.
Table 111. Aggregation Study of
1L in THF at 26 O C . ~~
13.2 6.56 3.06 0.769 0.568 a
21.38 21.36 21.36 21.32 21.30
BnMP uased as indicator.
silicon-stabilized carbanions do not shift significantly relative to the anions of the parent hydrocarbons, and a change in substitution on the silicon from methyl to tertbutyl does not substantially change the spectra of the cesium and lithium salts. In recent papers, this laboratory has presented revised indicator scalesof transmetalation potentials in THF with cesium and lithium as the gegenions.ls-lg The arbitrary reference point is that of fluorene at pK, = 22.90 per hydrogen for the two scales. Employing the same technique as before, the ion pair acidity differences (ApK's) of these salts were determined by measuring the equilibrium (1)spectroscopically.
RH + In-M+ R-M+ + InH (1) Table I1 lists the ion pair acidity results of the silicon compoundswith different indicators. The concentrations uaed are in the range of 1Wto 1V M,and the equilibrium constants do not change with concentration. An example is shown in Table I11 in which the PKLi of 1 relative to g-benzyl-9H-benzo[deflfluorene(BnMP, benzylmethylenephenanthrene) remains the same within experimental error over a 20-fold range in concentration. The apparent slightchangewith concentration correspondsto an average degree of aggregation of 1.05.22 The carbanion solutions (22) Knufman,M. J.;Streitwieeer,A., Jr. J.Am. Clrem. Soe. 1987,109, 8092.
1780 J. Org. Chem., Vol. 58, No. 7, 1993
Streitwieser et al.
Table IV. Rater of Equilibration of Cerium Saltr of Benzyltrimethylrilane and pMethylbipheny1 in THF at 26 O C
Table V. Overlapping Thermodynamic Mearurementsa
Li+ 4.60 x 10-2
2.71 X lo-'
61.6" 67.4b
3.13 X
4.32 X 10-5
402
[PhCHCeTMSl (M)
[BiPhCHs] (M)
5.25 X 10-5
0.111
4.37 x 10-5
0.668
41.Bb avg:
133 145 128 134 135
106kob (8-l)
10%-2 (M-ls-l)
9.27' 10.4d 5.13' 5.29" avg:
8.35 9.38 7.69 7.95 8.34
"From decrease in absorbance at 472 nm. bFrom increase in absorbance at 372 nm. From decrease in absorbance at 372 nm. From increase in absorbance at 472 nm.
were stable for several hours in the glovebox. The major source of error is most likely the extinction coefficient values used to convert absorbance readings to concentrations; unless noted, the accuracy of the pK values listed in Table I1 is about fO.l pK unit. At high pK's the rates of proton transfer are slow. For 4 the equilibria required about 8 h and permitted kinetic measurements. These were made from both directions so that the equilibrium constant could be obtained from kdk-2. The results for the equilibriumof 4 with the cesium salt ofp-methylbiphenylare summarized in Table IV. The rate ratio of 16 corresponds to ApK = -1.2 from which the pK of 4 is 37.3, in good agreement with the equilibrium results. Less extensive study was made of 5, but results with two indicators gave concordant results; the cesium salt of 5 showed spectral changes indicatingdecomposition on long standing (several hours). The pKCsITHF value obtained here for 4, 37.5, does not agree well with the PKLijTHF value, 38.9, derived by Miah and Fraser." The latter value comes from NMR measurements of equilibria with tetramethylpiperidine. The lithium pK, if anything, should be lower than the cesium value because the lithium salt has been shown to be a CIP.21 It seems likely that the lithium amide pK's are in error because known aggregation effects were not taken into account. The results show significant contrasts in the carbanion stabilizationof silicon. The trimethylsilylgroup stabilizes fluorenyl anion by only about 1unit both for cesium and lithium ion pairs (PKLiITHF and ~ K C ~ I T=:H 22.90 F for fluorene; PKLiITHF = 21.67, PKCqTHF = 21.32 for 1). Stabilization by the tert-butyldimethylsilyl group is somewhatgreater, about 2 pK units. A similar comparison of the acidities of 4-methylbiphenyl (pKcsITHF = 38.73) and 3 (pKcSITHF= 35.36) demonstrates that the acidityenhancing effect of the tert-butyldimethylsilyl group is now higher at 3.4 pK units. For toluene, ~ K C ~ I Thas HF been shown to be 40.9.19 The pK of 37.5 for 4 also shows stabilization by the TMS group of 3.4 pK units. Similarly, incorporationof a secondTMS group into the methyl group of toluene results in a further decrease in pK by 3.4 units; these results show that the effect of TMS groups is approximately additive. The differencein the magnitude of the stabilization is probably associated with the relative degrees of delocalization of charge in the two cases. In the benzyl and biphenylylmethyl anion there is effectively
cs+
2 BnMP BnMP 2 9-BnMP 9-BnFl
BA 2 BA 9-MeFl
2 9-MeFl
-0.6 -1.5 -2.1 (-2.9)e 2.7 -1.4 1.3 (1.7)e
-1.6 0.2 -1.4 (-1.46)e 1.1 -1.3 -0.2 (4.8).
For the reaction RIM + RzH RIH + RzM. See Table 11 for abbreviations. In kcal mol-'. In eu correctedfor symmetry. e Direct measurement in parentheses.
greater charge at the site of substitution and the stabilization effect of the silyl group is greater. This generalization was extended to other carbanions expected to be relatively localized. Measurementa were made of the cesium ion pair acidities of 6,7, and 8. These measurements were done with the single indicator technique, but recent improvements in this technique have provided results of good precision.23 The cesium salt of l,&dithiane has been shown to be monomeric in dilute THF solution.23 It seems probable that the cesium salt of 2-(trimethylsilyl)-l,3-dithiane,6C, is also monomeric; the TMS group in 6 causes a decrease in pK by about 3 units. Possible aggregation of the cesium salts of 7 and 8 could not be determined. It seems likely by comparison with the dithianes that 8C is monomeric but 7C could well be aggregated. An incremental effect of 3-3.5 pK unite per TMS group provides an extrapolation from 8C to the pKc.ITm of methane of about 47; this value compares with the recent estimate of 48-49 derived from dithianea.* Note that the effect of a TMS group is significantly lese thanthat of phenyl. Triphenylmethane(pKcl/m = 31.26) is substantiallymore acidicthan8in contrast to the relative kinetic acidities.I3 This comparison emphasizes that Bransted correlations are different for localized and delocalized carbon acids. Thermodynamic Data. The pK's of 2 were determined at several temperatures over the range of 25 to -20 OC. AHo and ASo for the ion pair equilibria were then determined via Van't Hoff plots. The resulta recorded in Table V show satisfactory agreement and indicate that the pKMIm values and the derived thermodynamic quantities have no important systematic errors. The values given are expected to be correct to about 0.6 kcal mol-' and 1.0 eu for AHo and ASo, respectively.16 An alternative approach is to analyze the thermodynamic parameters on the basis of a scale anchored on a single indicator. For the cesium runs, fluorene is a convenient standard;however,fluorenyllithiumis a mixture of contactand solvent-separated ion pairs at room temperaturemand 9-phenylfluoreneserves as a more convenient reference." In Table VI the relative enthalpy and entropy data for the salts of 2 are listed together with the values of some indicators. The enthalpy changes show the carbanionstabilizing effect of the silyl group-1.5-2 kcal mol-' relative to methyl or benzyl-but the entropy changes are comparable for the three substituents. The resulta also confirm the SSIP nature of 2L and the CIP nature of 2C.u Conductivity and Free Ion Acidity. Although the alkali metal salts of the silyl-stabilized carbanions exist predominantly as ion pairs in THF, the free ions are (23)Xie, L.;Bore, D. A.; Streitwiewr,A. J. Or#.Chem, 1092,57,498690. (24) Gronert, S.;Streitwieser,A. J. Am. Chem. Soc. 1988,110,283642.
J. Org. Chem., Vol. 58, No. 7, 1983 1781
Ion Pair Carbon Acidities of Some Silanes
Scheme I
Table VI. Comparihon of Thermodynamic Data in THF
Li+O
CS+*
RHc
m o d
Asoe
RHc
M o d
S o c
BPhFl BA BnMP 2
(0.0) 3.3 6.2 4.3
(0.0)
F1 9-MeFl 9-BnFl 2
(0.0) 0.8 0.2 -1.5
(0.0)
-0.3 4.2 2.8
+
*
RH
+
InM
KM~THF = RM
+
InH
4.5 5.6 4.4
For the reaction RH 9-PhFlLi RLi + 9-PhFl. Data for the indicators are in ref 17. For the reaction RH FlCs RCs + FlH. Data for the indicators are in ref 16. See Table I1 for abbreviations. In kcal mol-'. e In eu corrected for symmetry.
*
+
*
*
Table VII. Dissociation Constants for Lithium and Cesium Salts of 1-3 and Free Ion Acidities in THF at 26 OC compd
1L 2L
2c 3c
l@Kd (M) 1500 2300 7.8 10.6
pK(F1)' 20.9 19.8 19.8 34.8
0 On a per hydrogen basis. Defined as pK(F1) = pK(1P) - log &/&where& the diesociation constantof the standard,fluorene. = 1.5 X 1 p M for the lithium sdt, = For the cesium salt, 0.69 X le M.
important in the dilute solution reactions of delocalized carbanions because they are generally more reactive.% Moreover, it is useful to compare the silyl group stabilization effect with frse carbanions. Thus we measured the dissociation equilibrium (2)for cesium and lithium ion pairs by the conductometric method described previously.18 RM T=! R- + M+ (2) The dissociationconstant Kd was derived using the Fuoes equation% (3)
FJ& + fCAJFK,&'
(3) where F is the Fuoss term, A is the equivalentconductance, f is the mean activity coefficient, and A0 is the equivalent conductance at infinite dilution. As in the previouswork,18 F was calculated according to Fuoss and is equal to 0.91 to 0.96 for the lithium runsand unity in the cesium runs; f was obtained from the Debye-HOckel equation; it was close to unity for cesium and about 0.84.9 for lithium. A0 values are often difficult to obtain with a high degree of precision. To be consistent with our previous work, it was assigned as 85 cm2 W1M-l for lithium salts and 120 cm2 n1M-*for cesium salts. From the conductivitydata, plots of FJA Venus f2CAJF showed good linearity with no curvature. These plots are shown as Figures S1-S3in the supplementary material. Kd was obtained from leastsquares analysis of the Fums plots. The conductivity measurement results in Table VI1 indicatethat the diesociationconstants of lithium ion pairs are 103 times greater than those of the cesium salts and are comparable to the values found previously for the indicator hydrocarbons18 and again confirm that the lithium salts are SSIP and the cesium salts are CIP. Solov'yanov et al. have reported kinetic studies on carbanionsincluding the 9-fluorenyltrimethylailanesystem in dimethoxyethane(DME).% It is interesting to compare the properties of the silyl-stabilized carbanions in the differentsolvents. T w o features are consideredhere. Their (25) Fuoea,R. M.; Accaecina,F.Electrolytic Conductance;Intarncience Publishers Inc.: New York, 1959. (26) Solov'yanov, A. A.; Bebtrkaya, I. P.Zh.Org.Khim. ISM, 19,1822 (Engl. "ranel. p 1592).
spectral results for 1L in DME (Ama 362.5nm, c = 12 400) are similar to ours in THF (A, 364 nm, c = 12000). Similarly,their value for Kd, 0.9 X 105M, for 1L in DME is closeto that in THF, 1.5 X 106 M. That is, the properties of the SSIP are comparable in both solvents. There are significantdifferences, however, for the CIP cesium dts; for lC,c is higher in DME, 12 400,compared to 9760 in THF, although Am's are essentially the same, 357 and 356 nm, respectively. The Russian group reported Kd = 2.6 X le7M for 1C in DME. Although 1C is stable in THF for several hours our conductivity studies of this compoundwere of limited precision;however,we estimate a Kd of less than 7 X 10-8 M, comparable to that for 2C (Table VII). These results suggest that the bidentate nature of DME providesgreater solvationfor cesiumcation and greater ion pair dissociation compared to THF. The dissociation constants can be used to derive the relative free ion acidities of the silicon compounds. The dissociation constants of the silyl-substituted ion paire are somewhat higher than those of the respective fluorene salts.18 Thus, the relative ionic acidities of the silyl compounds are greater than those of the ion pairs. From the dissociation constants Kd the free ion acidities were derived from the ion pair pK assignments by the thermodynamic cycle shown in Scheme I. The pK(F1) is also referenced to fluorene at pK(F1) = 22.90 per hydrogen. The results are summarized in Table VII. The pK(F1) values of 2 calculated from the lithium and cesium acidity scales are in agreement and indicate the absence of important systematic errors. The free ion pK of 1 derived from the lithium salt, 20.9,is in moderate agreement with the DMSO pK reported by Bordwell.lS From the free ion acidities, the intrinsic silyl group stabilization effect can be evaluated quantitatively: trimethylsilyl and tert-butyldimethylsilyl groups lower the pK(F1) of fluorene by 2 and 3 units, respectively. These values are somewhat lower than the 3-3.4 pK unit effect of the silyl group in the ion pair acidities of 3 and 4 and demonstrate further the moderate attenuating effect of silyl substitution with more delocalized carbanions. We note that the PKLi/THFvalues of 1 and 2 are 0.3 unit lower than the corresponding ~KC~ITHF, values. That is, relative to the hydrocarbon indicator standards, these silicon-stabilized carbanions are 0.3 unit more stable as SSIP than as CIP. Moreover, from the free ion acidity data, we note that the relative pK(F1) for 2 is even lower than PKLi/THF. The ~ K C ~ T(CW, H F PKIJTHF (SSIP),and pK(F1) for 1 are, respectively, 21.64,21.33,20.9,and for 2, 20.60, 20.28, 19.8. The same trend is apparent for 3; ~ K C ~ / Tand H F pK(F1) are 35.62 and 34.8,respectively. Although the decreasesin the three kinds of pKvalues are not large, the trends point to the conclusion that the relativestability of the silyl-stabilizedcarbanionsare cation dependent;the silicon-stabilizingeffect increasesfrom CIP
1782 J. Org. Chem., Vol. 58, No.7, 1993
to SSIP to free ion. The effect may be determined by the size of the silyl group compared to the hydrocarbon standard. The larger group results in an increase in carbanion-cation distance and less ion pair stabilization. However, analysis is complicated by electrostatic interactions involving polarization at silicon and the highly polar nature of silicon-carbon bonds.27 It is interesting that the same type of effect is seen in the ab initio calculationsof Schleyeret al.9 At the 3-21G(*) level, SiH3 stabilizes methyl anion by 25 kcal mol-'. In contrast, in methyllithium and methyl", SiH3 substitution providea stabilization of only s11kcal mol-'. The calculations show the same trend for PH2 and SH substitution and suggest that comparable results with related ion pair and ionic acidities would apply to such Substituents. Silyl Group Transfer. Many studies of substitution at silicon by nucleophiles have been carried out as summarizedin an earlymonograph by Sommer= and more recently by Corriu.29 It is well-known that oxygen and halogen bases and nucleophilesprefer to attack at silicon, whereas carbonand nitrogen bases and nucleophilesprefer to attack at hydrogen. However, because Si-C bonds are polarized with a high positive charge on the Si atom,27in some cases carbanions do attack at silicon as nucleophiles to substitute the silyl group and result in a new carbanion. For example, attempts to deprotonate (methoxymethylltrimethylsilane, MeaSiCHzOMe, using n-butyllithium resulted in attack at silicon and subsequent cleavage of the S i 4 bond.30 Perozzi et showed that PhLi can attack a neutral four-coordinatesiliconate to result in fivecoordinate siliconate. In our study, it was found that in some cases the substitution at silicon by the indicator carbanionscompeteswith proton transfer. The silylgroup functions as a 'big" proton to transfer from one carbanion to another. This process depends on the basicity of the base and the leaving group, the metal gegenion and the substituent groups on silicon and is usually much slower than proton transfer; for example, the long equilibria involving 4 and the corresponding kinetic studies were not affected by silyl group transfer. Such transfer was demonstrated with the use of 9,9-bis(trimethylsilyl)fluorene, 9. 9-Methylfluorenylcesium (~KC~ITHF = 22.3) or 9-tert-butylfluorenylcesium(~KC./THF = 24.4) substitutes the trimethylsilyl group in 9 only very slowly. The cesium salt of 9-phenyl-10,10-dimethyldihydroanthracene (PDDA, PKC'ITHF= 28.1) attacks at the silicon in 9 at faster rate and reaches equilibrium, eq 4, with log K = 0.2-0.9.
9
p-Biphenylylmethylcesium (pKc.~m,= 38.71, on the other hand, reacts with 9 completely at a rapid rate. The nature of the ion pairing plays a role in silyl group transfer. In contrast to the reactions with cesium bases (27) Gronert, S.; Glaser, R.; Sbitwieaer, A., Jr. J. Am. Chem. Soc. 1989,111,3111. (28) Sommer, L. H. Stereochemistry, Mechaniem and Silicon; McGraw-Hill: New York, 1985. (29) Corriu. R. J. P.; Guerin,C. Adu. Orgonomet. Chem. 1982,20,266. (30)Magnua, P.; Roy, G. J. Chem. Soc., Chem. Commun. 1979,822. (31) P e d , E.F.;Michalak, R. 5.; Figuly, G. D.;Steveneon, W. H., HI;Dw,D.; Roa, M.R.; Martin, J. C. J. Org. Chem. 1981,46, 1049.
Streitwieeer et al. discussed above, the lithium bases made from indicators with PKLiITHF in the range from 19 to 33 do not undergo the substitution reaction at silicon. Thus, silyl group transfer is more facile for CIPs than for SSTPs. Sterichindranceto substitution at siliconis pronounced. This effect has been found in the substitution at silicon by oxygen and halogen nucleophiles." It was also found for the substitution by carbanions as nucleophilea in this study. When the substituents on siliconare changed from trimethyl to tert-butyldimethyl, the silyl group transfer is completely blocked even with cesium salts.
Conclusion This study has led to a quantitative understanding of silylstabilizationin carbanions. Relativeto hydrocarbons, the intrinsic stabilization (free ion stabilization) is about 2-3 pK unite for delocalized carbon acids but is somewhat greater for localized carbanion systems. The stabilization decreasesfrom free carbanionsto organometallicion pairs and is greater for SSIP than for CIP.
Experimental Section General. Glassware was dried in an oven at 160 O C for at least 4 h prior to use. T h e meaeurementa of trammetalation equilibrium conatantawere made in a gloveboxfiied with Argon. W-Vie spectra were recorded on an IBM 9431 W-via spectrophotometer or on a ShimadzuUV-2101PCspectrophotometer equipped with fiber optics cables connected to a thermostatted cell holder in the glovebox. Conatant temperaturewan obtained by using a Nealab Model RTE-8DD temperature controller for room temperatureand higher and by Neslab Model ULT-ODD for lower temperat" using 96 94 ethanolaa the coolant. Proton nuclear magnetic resonance ('H NMR) spectra were conducted on UCB-250 (a superconducting 2WMHZ instrument),Bruker AM~,andBrulrerAM-M)O(FTNMR)spectrometereoperatine in the FT mode. Gas-liquid chromatography (GLC) was performed on a HewlettPackard6880gas chromatograph using a 6-ftx Isin. OV-101column. Melting pointa (Pyrexcapillary) were determined on a Buchi melting point apparatus and are
uncorrected.
Materialo. Commercial materiala 4,6,7, and 8 were dried and distilled. Compounds 1,- 5,s and P a r e described in the literature. The indicatore used were available from our previous studies. The purity of the compounds was monitored by combinatione of *HNMR, GLC, mp, and elemental analyeis. ~Fluorenyl-~~butyldimethylrilone,2. ToaO oCsolution of 2.7 g (0.016 mol) of fluorene in 20 mL of THF WM added over 30min8.7 mL of 2.3 M n-butyllithium(inhexane). The mdting red solution was added via a cannula to a flaek containing magnesium bromide which was made from 0.6 g (0.020 mol) of magnesium turninge and 3.7 g (0.020 mol) of 1,2-dibromoethane in 26 mLof THF. After 1h of stirring,to the fluorenylmagneeium bromide solution was added over 1h 4.0 g (0.026 mol) of tertb u t y l d i m e t h y l c h e in 10 mL of hexamethylphosphoric triamide (HMPA) and 10 mL of THF. The solution waa stirred for au additional8 h at 50 O C . The reactionmixture was extracted with 1M HC1 and 3 X Ha0 to give a brown ether solution. The solution was dried over anhydrous magnesium sulfate, and the solvent WM removed by rotary evaporation under reduced preeeure to give a brown oil. Fractional aublimation under vacuum produced white crystab as the second fraction: mp 7879 O C ; 'HN M R (CDCL) S 7.7-7.1 (m, 8 H), 3.8 (a, 1H), 0.23 (s, 9 H),0.18 (8, 6 H). Anal. CalcdforCl&&i: C,81.40,H,8.57. Found C,81.61; H,8.62. (32) Gilman,H.; Benkewr, R. A.; Dum, G. E. J. Am. Chem.SOC. 1960, 70,1689. (33) Eabom, C.; Shaw, R. A. J. Chem. Soc. 1966,1420. (34) Bey, A. E.; Weyenberg, D.R. J. Org. Chem. lwb, 91,2086. (35) Merker, R. L.;Scott, M.J. J. Am. Chem. SOC.1 W , 86, 2 u 9 .
J. Org. Chem., Vol. 58, No. 7, 1993 1785
Ion Pair Carbon Acidities of Some Silanes (4-Biphenyl~lmethyl~-tert-butyldimethylsilane, 3. To a 0 "Csolution of 2.7 g of 4-methylbiphenyl(O.O16 mol) in 20 mL of THF was added over 20 min 8.7 mL of 2.3 M n-butyllithium in hexane. The Cbiphenylylmethyllithiumsolution ivaa slowly added via a cannula to a flask containii magnesium bromide which was made from 0.5 g of magnesium turnings (0.020 mol) and 3.7 g of 1,2-dibromoethane (0.020 mol) in 25 mL of THF. After the solution was stirred for 1 h, 4.0 g of tert-butyldimethylchlorosilane (0.026 mol) in 10 mL of HMPA and 10 mL of THF was added to the solution. The solution was stirred for 5 h at room temperature and quenched with 0.5 M HC1. The mixture was extracted twice with diethyl ether, and the extracts were washed with 1 M HCl and water and then dried over anhydrousmagnesium sulfate. The solvent was removed under vacuum to give a yellow oil. Fractional sublimation under vacuum produced white crystals as the second fraction, mp 85.5-86.0 "C; 'H NMR (CDCL) 6 7.67.1 (m, 9 HI, 2.1 ( ~ $H), 2 0.9 (8, 9 HI, -0.1 (8, 6 HI. Anal. Calcd for ClgHaeSi: C, 80.83; H, 9.22. Found C, 80.62; H, 9.22. Purification of THF. Carefully dried THF is especially importantin the single indicator studies. Fisher Scientificreagent grade THF was distilled from lithium aluminum hydride or from sodium-benzophenone into an oven-dried Schlenk flask containing a magnetic stirring bar. The flask was attached to a vacuum line and the content degassed by at least three freezepump-thaw cycles. It was then taken into the glovebox, and several drops of sodium-potassium alloy were added. The solution was stirred in the glovebox until a blue color persisted. The flask was then taken out of the glovebox and attached to the vacuum line, and the contents were degassed by three freezepump-thaw cycles. During the degassing process, the blue color of THF solvated electron might disappear due to freezing. In such cases, the solution was stirred after degassing until a blue color reappeared, and THF was vacuum transferred to a long neck receiving flask which had been oven- and flame-dried. The contenta were again degassed and taken into the glovebox, and approximately 20 g of 3-A molecule sieves was added for every 250 mL of THF solvent. The solvent was allowed to stand over sieves for at least 2 daysin a quartz flask prior to use. We estimate the water content to be on the order of 10-8 M. Acidity Measurements. General procedures of the double indicator method and singleindicator method have been described previously.29 The single indicator method was further improved by using quartz microbeakers. Earlier work with the double indicator method made use only of the .A, of the substrate and indicator salts. In more recent determinations, the entire spectrum was used. To deconvolute the mixed spectra, we have used a linear least-squares computerprogram that calculates the based on absorbance of both anions at their corresponding A, the inputs of the standard spectrum of each individual anion. The details of the linear least-squares method will be described in a separate paper.% From our experience, this linear leastsquares method is preferred to the general method of solving two linear equations for the absorbance at two wavelengths. An example is shown as Figure S4 in the supplementary material. The extinction coefficients for most of the indicator and substrate anions were measured by titrating to the end point of a known amount of neutral in a known volume of solvent with a strong base (diphenylmethylcesium for titrating compounds with pK.'s lower than 30 and cumylcesium for thoee with pK.'s higher than 30). For anions whose spectra are not interfered with by the spectrum of the base, the end point was taken when further addition of the base caused no further increase in the absorbance of the anions. For thoeeanions whoee spectraoverlap with the spectrum of the base, an excess amount of the base was added (overtitrated) and the mixed spectra of the anionsand the base were then deconvoluted by the linear least-squares method used for the double indicator method to give the end point. For compounds with pK.'s higher than 36, the deprotonation by cumylcesium becomes very slow. The measurement of extinction coefficients by titration methoda described above involve experimental difficulties with greater uncertainty. In (38)Krom, J. A.; Petty,J.T.; Streitwieaer,A.Manwcriptin preparation. (37)Streitwiener,A.; White, J.; Xie, L. Manuscript in preparation.
2,
1.5
I
i 350
400
450
Mo
5%
Wavelength (nm)
Figure 1. Repeated scans in a kinetic experimentof measthe forward rate constant of 4 with the cesium salt ofp-MB (A= 472 m) at 25 OC. The .A, of 4C = 372 nm.
A (472 om) 1.1
0
40
60
Time (minutes)
110
160
Figure 2. Absorbance of reactant anion and product anion as a function of time in the kinetic experiment of Figure 1. such casea, we have used an alternative method, which proved more reliable and internally consistent. The following methad was used to determinethe extinction coefficientsof 4 C 3-10 mg of neutral compound (so that the rate of deprotonation was convenient to follow) was placed in a microbeaker and added into a cellcontaining THF. A base-lie correctionwas performed on the cell. An appropriate amount of cumylcesium solution was added so that the absorbance of cumylcesium at A, 346 nm reached about 1.0 absorbance units. Repeated scans were recorded every 3-6 min. The seriea of mixed spectra of cumylceaium and the substrate anion were then deconvoluted usingthe linear least-squares method,and the abeorbauce at 346 nm and at the A, of substrate anion was obtained. The ratio of the decrease in the absorbance at 346 nm to the increase in abeorhce at.A, of the substrate anion for each pair of succBBBivb spectra thus yields the ratio of the extinction coefficients of cumylceeium (c = 17 OOO) to that of substrate anion. Several ratios were obtained and averaged. This method was found to give rapid and reliable extinction coefficients. Their internal consistency was checked, and the uncertainty of extinction coefficients was estimated to be 5-8%. Thermodynamic constants were obtained by determiningthe pICs at 5-10 "C intervals over a temperature range of 25 to -20 "C. Linear least-squares plota of log K va 1/T were obtained and ASowere determined. The thermodynamic from which MHO resulta are summarized in Tables V and VI. Conductance measurements were done within the glovebox. Conductancewas measured with an ESI Model 2110 VideoBridge operating in the parallel circuit mode usinga procedure detailed previously.18 Examples of the derived Fuoee plots are shown in Figures 51-53 of the supplementary material. Derived dhociation constants are summarized in Table Vn. Kinetics. For benzyltrimethylsilane, 4, p-tolyl phenyl amide and p-methylbiphenyl were used as indicators. Thee0 transmetalation reactions were slow and required approximately 8 h
1784 J. Org. Chem., Vol. 58, No.7, 1993 before equilibriumwaa reached. For both the forward and reverse reactions, a large exceaa of neutral compounda @-MBand 4) waa used and the reactions were found to be pseudo-firaborder in the cesium aalt. The reactions were conveniently monitored by successive scana from 550 to 350 nm at a constant time interval (Figure 1). The cesium aalt ofp-MB haa A, 472 nm; 4C haa A, 372 nm. Note that the sharp iaoabestic point indicates that silyl group tranafer is not important on thia time scale. The mixed spectra were deconvolutedusing the linear leaabaquares method, were plotted against time. An and the absorbance at both A, example of such plots ia shown in Figure 2 for the reaction in Figure 1. The data were then fitted against the usual equation for f i b o r d e r kinetics using nonlinear leaat-squares. Division of the paeudo-fit-order rate constantby the concentrationyielda the second-orderrate constant. Sincethe absorbancesof readant anion and product anion were both followed at the same time, two rate constants were obtained for each run.
Streitwieser et al. Silyl Group Tranrfer Studier. The U V 4 akorption spectrum was taken of the lithium or ceaium aalt of the appropriateindicator. A known amount of 9,9-bie(trimethybilyl)fluorene, 9, was added and the extent of silyl group tramfer WBB determined from the decrease in the indicator abaorption and the appearance of the correapondingaalt of 1. Only qualitative obaervations were made.
Acknowledgment. This research was supported in part by NIH grant no. GM-30369.We thank Dr.James C. Ciula for preliminary experiments. Supplementary Material Available: FUMEplots and an experimentalspectrum (5 pages). "hh material ia contained in libraries on microfiche, immediately follows thia article in the microfilm version of the journal, and can be ordered from the ACS; aee any current masthead page for ordering information.
